Comparative linguistic analysis of noun affixal derivatives

Abstract

The verb-noun pairings in the Princeton WordNet were subjected to a morphosemantic analysis. The findings are shown in the standoff file, which has pairs annotated with a set of 14 semantic connections. We detected the affixes, automatically differentiated between zero-derivation and affixal derivation in the data, and manually verified the outcomes. The findings indicate that an affix predominates in the creation of new words for each semantic relation. However we are unable to discuss their specificity with regard to such a relation. Additionally, for each semantic connection, some verb-noun semantic prime pairings are better represented than others, leading to the emergence of various semantic clusters (in the form of WordNet subtrees). In order to capture finer regularities in the derivation process as represented in the semantic properties of the words involved and as reflected in the structure of the lexicon, we therefore employ a large-scale data-driven linguistically motivated analysis made possible by the rich derivational and morphosemantic description in WordNet. [1:42]